I’ll be up front. This one won’t wrap up in a tidy bow.
Sometimes, we only glimpse the future, and we can’t quite understand it completely. At least, that’s what I think might have happened to me when I read Steve Staude’s extremely nerdy extremely/great piece on forecasting strikeout rates for different pitcher/batter matchups. Seriously, the piece asks a simple question, but the method and the answer has implications all over baseball.
Instead of asking what player X would do against pitcher Y, Staude matched all batters with certain strikeout rates against all pitchers with certain strikeout rates to see what happened in those at-bats. It’s worth reading, but there’s a matrix there which says that if a pitcher with a 15% strikeout rate meets a hitter with a 15% strikeout rate, the batter will strike out 11.4% of the time. It’s less than 15% because the league strikes out 20% of the time, so both the hitters and pitchers are used to more strikeout punch from their competitor. Basically, a 15% strikeout pitcher strikes 15% of batters with a 20% strikeout rate because that’s what the league is doing right now.
There’s not a ton that we can take away for fantasy baseball immediately. At least not for yearly leagues.
But daily fantasy players might be interested. And they can take something away right away. Perhaps that pitchers and hitters have an equal effect on strikeout rates. If you’re looking at a pitcher with a high strikeout rate against a team with a low strikeout rate, you may want to keep looking at other possible starters. If you’re looking for strikeouts for your starting pitcher, maybe look for a matchup with a team other than the Giants, Tigers, and Orioles, who are the top three teams at avoiding the strikeout (all <17.1%). Maybe that’s a no-brainer, but maybe not: If you were throwing Cliff Lee against the Giants, you might not check this stat at all. Maybe you shouldn’t. But Doug Fister against the Orioles might be a bad idea.
And the converse is true: you might be able to take advantage of the Astros, Braves, and Mets. Scratch that — you definitely already knew about that one.
But the really fantastic thing is that this model can be used to analyze any one-on-one matchup to get an idea of who influences that confrontation more than the other.
Fly-ball hitters versus ground-ball pitchers? Junk throwers against fastball hitters? Stolen bases? Is it the catcher, the pitcher, or the runner? (Unsure if the model cant handle three different, but we’re riffing here.) Could we attack framing with this — good framer, wild pitcher?
The possibilities are exciting. At some point, once these many different possibilities are sussed, you might be able to use a matchup machine to throw in a pitcher and a hitter, or two teams, to spit out some likelihoods. You could throw a few pitchers in the machine to see which matchup is best for streaming, even. All sorts of fun fantasy uses.
Of course, it could also be used to replace the antiquated batter-vs.-pitcher data that some managers still use, and it could upgrade in-dugout strategy pretty fiercely, but that’s “real” baseball. Whatever.